The Five Eyes statement is notable for its urgency: it says frontier AI will reshape offensive and defensive cyber capabilities on a timeline of months, not years. This is a rare joint call to action from the alliance's cybersecurity agencies.
cybersecurityTuesday, June 30, 2026
Five Eyes warns on AI threats, Mustang Panda strikes India
The Five Eyes alliance issued an urgent warning that frontier AI will transform cyber capabilities in months, not years. Meanwhile, Mustang Panda is abusing Zoho WorkDrive to target Indian government networks, and a critical Oracle EBS flaw is being actively exploited. The day's stories share a thread: adversaries are moving fast, and defenders need to match that pace.
Allied warnings and AI threats
The Five Eyes intelligence alliance issued a coordinated statement that sets the tone for the day: AI-driven cyber threats are accelerating faster than expected.
Mozilla's 0DIN team demonstrated a practical indirect prompt injection attack on AI coding agents like Claude Code, using a GitHub README to trigger a reverse shell. It's a concrete example of the kind of AI-driven threat the Five Eyes is warning about.
Active campaigns and exploits
While policymakers warn of future threats, attackers are already active today, Mustang Panda is abusing legitimate cloud services, and a critical Oracle EBS flaw is being exploited in the wild.
Mustang Panda is using Zoho WorkDrive as a covert C2 channel, hiding malicious traffic inside legitimate cloud activity makes detection harder. The group is targeting Indian government and hydropower organizations with spear-phishing and sideloading.
CVE-2026-46817 is a critical Oracle E-Business Suite flaw (CVSS 9.8) being actively exploited in the wild. Oracle patched it in the latest CPU, but Defused Cyber observed exploitation on honeypots, meaning unpatched systems are at immediate risk.
Also today3
Incorporating the Cyberspace Domain: How Russia and China Exploit Asymmetric Advantages in Great Power Competition - Modern War Instituteshare.google
NeuroSploit V353 – The Offensive AI That Thinks, Validates, And Strikes Autonomously + Video - Undercodeundercodetesting.com
SAP Security In 2026: A Zero-Day Survival Guide For The Modern Enterprise + Video - Undercode Testingundercodetesting.com
More roundups that day
ai & machine learning roundupAI faces fool people, Codex goes local
AI spots heart risks in routine ECGs
Disability rights and voting access dominate
AI spots heart risk in routine ECGs

Smartphone ban debate heats up in England
